From owner-freebsd-isp Tue Sep 16 10:16:15 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.7/8.8.7) id KAA06658 for isp-outgoing; Tue, 16 Sep 1997 10:16:15 -0700 (PDT) Received: from indigo.ie (aoife.indigo.ie [194.125.133.9]) by hub.freebsd.org (8.8.7/8.8.7) with ESMTP id KAA06652 for ; Tue, 16 Sep 1997 10:16:12 -0700 (PDT) Received: from indigo.ie (localhost [127.0.0.1]) by indigo.ie (8.8.5/8.8.5/INDIGO-HUB) with ESMTP id SAA25378; Tue, 16 Sep 1997 18:15:57 +0100 (BST) Message-Id: <199709161715.SAA25378@indigo.ie> To: Steve Ames Cc: freebsd-isp@freebsd.org Subject: Re: best news server config In-reply-to: Message from Steve Ames dated today at 09:37. From: Alan Judge Date: Tue, 16 Sep 1997 18:15:57 +0100 Sender: owner-freebsd-isp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk We have a PPro 200, with 256MB of RAM. We have eight 4GB Quantum Atlas II drives spread across two channels of a 3940UW. (Disks are used as: one system/history/logs, one overview, two stripped for alt/binaries, and four stripped for the rest of spool.) CPU and memory are fine. I/O mostly OK, but probably need to spread the history/log load out. We're running INN-1.5.1 with a few patches. Fairly reliable, except for an intermittent problem that causes the active file to get corrupted (one extra space at the very end; innd doesn't care; nnrpd does), and regular messages about .overview files having 'bad format'. Since the last hike in news flows (up to around 700K articles/day), we're only barely keeping up. Though I think that network congestion (due to pulling feeds across transatlantic links) is an issue for us, we don't have much spare capacity. The inntimer stats show little free at busy times. About half the time is spent writing articles, the rest scattered. At full whack, our system processes a max of about 3000 article writes in 5 minutes with no idle noticable. That's only 860K articles in a day. Nothing like the spare I planned for when I set up the machine and news was flowing at 300K/day. Your system is very likely maxed out in I/O. Have a look at iostat or systat to see how many tps your disks are taking. You may well need more memory. INN likes lots. Consider increasing NBUF as well if you get more memory. I'd be interested to hear what other people are experiencing. We're thinking about installing a Diablo front end machine to smooth the transatlantic flow and do some spam deletion before things get to INN. -- Alan Judge Phone: +353-1-6046901 Indigo Network Operations Centre Fax: +353-1-6046948